Robin Support
Practical guides, setup steps, and troubleshooting for running Robin on your Windows PC.
Getting Started
Platform notes and initial setup
Google Workspace
Connecting Gmail, Calendar, and Tasks
Local AI (Ollama)
Running models on your hardware
Voice & Audio
Microphone setup and wake word
1. Getting Started
Robin is a desktop application designed specifically for the Windows environment.
Platform & Prerequisites
- Operating System: Windows.
- Audio Input: A working microphone and speaker output if you wish to use voice interaction.
- Internet Access: Local Ollama model inference itself can run without internet access. Features that use Google services, cloud models, or other online services still require a network connection.
2. Google Workspace Setup
Robin connects to Gmail, Google Calendar, Google Tasks, and Google Contacts through standard OAuth authorization.
- Open Settings: In the Robin application, navigate to the Settings panel.
- Connect Google Account: Click "Connect Google Account" to open the sign-in page in your default web browser.
- Authorize Access: Review the requested scopes (Gmail drafts and messages, Calendar events, Tasks, and Contacts) and click Allow.
- Secure Storage: Robin receives OAuth tokens and stores them in your Windows Credential Manager under the service name
Robin.
3. Local AI (Ollama)
If you want all assistant reasoning to run on your local PC, configure Robin to use Ollama:
- Install Ollama: Install Ollama from ollama.com .
- Install a Supported Model: Download a supported model via Windows Terminal (for example:
ollama pull llama3:latest). - Configure in Robin: Ensure Ollama is running at
http://localhost:11434. In Robin Settings under "AI Engine", select "Ollama", discover available models, and choose your downloaded model from the list.
4. Cloud AI (Google Gemini)
If you prefer cloud model reasoning, Robin supports Google Gemini:
- API Key: Generate an API key from Google AI Studio.
- Enter Key: In Robin Settings, select "Cloud (Gemini)" and enter your API key.
- Model: Robin connects directly to Gemini for assistant completions.
5. Voice & Hands-Free Interaction
Robin features an on-device streaming voice pipeline:
- Wake Word: Speak "Robin" clearly to activate listening.
- Microphone Access: Ensure Windows Settings > Privacy & security > Microphone allows desktop apps access.
- Audio Processing: Audio frames are processed in-memory locally on your CPU for wake-word spotting and speech recognition.
6. Scoped Workspace & Files
Robin allows you to summarize and reference files in its designated workspace directory:
C:\RobinWorkspace
Robin does not search outside this folder. You can configure a different directory path in Robin Settings.
7. Action Approvals (Human-in-the-Loop)
Robin requires explicit human review before performing mutating actions.
- Reading Data: Checking your calendar, listing tasks, or reading workspace files does not require approval gates.
- Modifying Data: Sending an email, altering a calendar event, or updating tasks presents an approval card with action parameters.
- Confirmation: Robin only executes when you click "Approve". Clicking "Cancel" discards the proposed action.
8. Common Troubleshooting
Q: Ollama connection error
Verify that Ollama is running on your machine by opening a browser to http://localhost:11434. You should see "Ollama is running".
Q: Google authorization expired
If an authorization error occurs, open Robin Settings > Google Status, click "Disconnect", and re-connect your account to refresh tokens.
Q: Wake word not triggering
Check your microphone volume in Windows Settings > System > Sound > Input. Ensure the microphone is not muted and that Robin has microphone permission.
9. Support Contact
Official support mailboxes will be announced on this website prior to the public Windows release. Early preview testers may submit feedback through their designated communication channels.